Police STAR Fund Projects
Previous Police STAR Fund projects:
InterACT
Research to compare and contrast the accuracy of crime types against harm groups to reflect a ‘high harm’ approach to crime reduction in predictive mapping
Multimedia youth justice
Creating engaging, credible and effective behavioural-change resources for youth justice teams.
Forensic use and application of MALDI
Testing the operational value of new mass spectrometry methods of resolving and visualising fingermarks.
Wearables for wellbeing
Trialling the value of data analytics for tracking the wellbeing of staff.
LoRaWan tracking
Will establish a novel, secure method of asset tracking (e.g., agricultural equipment) for frontline officers.
Missing murdered
Identifying triggers that might indicate harm when taking missing person reports.
Automated redaction of case files
Developing a tool that investigators can use to accelerate redaction of sensitive and personal information.
Trafficking mediated online
Developing a risk assessment tool to enable better triage of ASW profiles.
Knife recognition AI
Developing a mobile-based app for instant knife recognition and comparison.
Serious misconduct prediction
Taking learning from Australia to test a data-driven model for serious misconduct prediction.
Digital road safety technology
Assessing the range and efficacy of new technologies that focus on behavioural change in at risk drivers.
Decision-making in evidence led prosecutions in domestic abuse cases
Examining decision making in relation to ELPs for domestic abuse and their effect on victims’ experience.
Building public trust (victim journey)
Developing a digital contact platform that allows victims to receive on-demand updates.
Machine learning - domestic homicide prediction tool
Enhancing domestic violence risk assessment methods with knowledge of factors predictive of homicide.
Compassionate policing - RASSO & VAWG (Project Odyssey)
Building a focused digital acquisition capability to support front line officers.
DaQIOA
Building a data quality dashboard and methods to correct and enhance data for serious violence reduction.
Traumatic brain injury
Testing how synthesising multiple data points can support investigation and harm reduction
Mark retrieval from knives
Create a semi-automated method for examining a knifes surface that can be operated by non-specialist
Cyber tabletop exercise
Creating remote-working focused training tool to improve knowledge of cyber-attack risk
Wi-Fi and cellular enabled VSS (CCTV) data extraction and analysis
Research the data storage and transmission structure of Wi-Fi and Cellular VSS/CCTV cameras to support forensic labs in their extraction and analysis of these devices.
Telematics: using machine learning to drive behavioural change
Create a platform which will automatically ingest telematics data and use machine learning to classify driver behaviour and drive an education loop to reduce accidents and associated costs.
A study of rib fractures in early childhood to aid forensic investigations of suspected child abuse
Study of paediatric rib cages using micro-CT and histology to assess the occurrence of fractures in infants that have been resuscitated but not likely to have been subjected to abuse. Findings from these control cases will be compared to suspected abuse cases to characterise injury patterns resulting from different mechanisms.
Meeting the policing demand of a digital world using an evidence-based approach
Establish an evidence-base for digital demand in policing and test a Digital Investigation Tool (DIT) using a randomised control trial.
VAWG reduction in Essex through the expansion of the Minerva tool
Evaluate Minerva to improve the way we identify and tackle VAWG incidents in public spaces.
Problem orientated policing local assessment resource (POPLAR)
Research, design and deliver a practical Problem-Oriented Policing (POP) Evaluation Tool for GMP Neighbourhood Policing (NP) teams, adopting a human-centred design approach to innovation.
Identifying and managing risk in non-convicted sex offence perpetrators
Develop, pilot, and evaluate a structured tool to a) guide police officers’ identification and risk assessment decision making of non-convicted serious sexual offence perpetrators, and b) better inform and encourage the use of different tactical disruption options available to tackle and reduce this type of offending.
“When all is tread and done” Identifying footwear uppers caught on camera as a new form of evidence
Develop a searchable database of footwear upper images under visible and infra-red light.
Project guardian
Targeted research in police pursuits to develop and evaluate alterative practices and adopt innovative new tactical options (utilising technology) to bring pursuits to a prompt and safe conclusion.
The prioritisation of potentially harmful subjects and the estimation of future risk
Evaluate outcomes of a predictive algorithm used to identify stalking and harassment (S&H) offenders who may pose high harm.
Kent and Essex ILAS trial
A 2-force trial of an intelligence automation application, focused on child exploitation and VAWG. The trial will be supported by academic validation.
NLP to detect threatening and abusive language toward victims.
Develop Natural Language Processing (NLP) models to reduce privacy intrusion when reviewing material extracted from a victim’s phone as part of an investigation.
CHISVal
Trial a new evidence-based template developed for objectively measuring the value of CHIS in context specific terms.
Targeting repeat sexual offending suspects using behavioural analysis
Explore, pilot, and review the use of BA and strategic, disruption-based policing options to effectively target and prevent repeat sexual offending. Several tools will be implemented in DCP to assess their utility in targeting repeat offending. A national, structured framework for the use of BA in policing will be developed
ANPR data quality
A discovery and proof of concept to develop quantitative measures of ANPR data quality and an associated dashboard. This would cover issues with ANPR cameras and an assessment of ANPR data feeds, as well as data quality impacts for operations and analytics. This builds on previous qualitative assessments.
Examining the use of evidence led prosecutions (ELPs) in domestic abuse cases
Explore the use of ELPs across four different police forces. To date, with the exception of the originally funded project, no research has examined how ELPs are working in practice, or the impact on victims.
101 voice assistant “Andi/Esra”
Build and pilot “Andi/Esra”, a 101 non-emergency call taking voice assistant to answer 101 call requests and deal with specific use-cases, reducing the service requests on call operators.
Community and commissioning data hub
Develop a multi-agency database and app that gives unified and locally-specific information on support services, programmes and diversionary activities to facilitate effective signposting and referrals, and enable better-informed future commissioning decisions.
Proof of concept to manage and share data relating to neurodivergent officers
Develop a proof of concept to collate and share information (only to authorised personnel) on workforce neurodiversity, to better support recruitment, promotion and operational delivery.
A GenAI platform for policing
Implementation of a Generative AI platform (including data connectivity and governance framework) to demonstrate how AI capabilities can be embedded effectively and securely into police forces, enabling innovation and scaling of AI across multiple and varied use cases.
Understanding and addressing fraud and cyber victimisation against children and young people
The research aims to develop our understanding of children and young people’s (CYP) experiences of fraud and cyber crime. We will combine insights from two large-scale surveys with victim interviews to better understand the prevalence, typology and impact of fraud and cyber crime on CYP. The research will result in a clear action plan to tackle fraud against CYP.
An evaluation of the new High Harm Unit in Dorset Police: Understanding the identification, assessment, and management of potentially harmful subjects.
This project will evaluate an innovative High Harm Unit in Dorset Police, established to improve identification, assessment, and management of perpetrators responsible for repeat domestic abuse or sexually harmful behaviour(s). The new intervention aims to enhance public protection by encouraging behaviour change or disruption over and above current investigative practice.
Evaluation and implementation of decision-support tool for identification and management of risk in non-convicted RASSO perpetrators
This project will implement and evaluate the structured decision-support tool that we have developed to identify and manage risk in non-convicted RASSO perpetrators. The tool has been piloted and developed in HDPP and will now be formally implemented in force and evaluated against other risk management initiatives.
Semi-automated identification of sex-trafficking victims and perpetrators
This project accelerates the identification and targeting of potential sex-trafficking activity. It will develop a new risk tool that will search large volumes of open-source data from Adult Service Websites using the Traffic Jam platform and auto-flag to officers the data of most concern for victim vulnerability and potential criminality.
Approaches to violence against sex workers in Gwent
This project evaluates policing responses to incidents of sexual violence reported to the police by sex workers in Gwent. In bringing together partners from policing, Cyfannol Women's Aid and academia, this work examines investigative approaches, community dynamics and processes of risk identification and management.
Sleep fatigue & recovery biometrics project
PWS in partnership with LJMU have developed the world’s first police specific, end to end biometric health & performance system which has the potential to positively impact the whole workforce. Comprising a highly accurate wearable device, a customised app & interactive educational platform the system has already been enhanced with user feedback.
Improving representation and inclusion in Serious Organised Crime policing roles
The NPCC SOC Portfolio is collaborating with the University of Portsmouth and four forces, to develop a mature understanding of diversity, embracing surface-level characteristics and expanding our understanding of hidden-level diversity. This will provide an evidence-base to define and remove unknown barriers, innovative approaches to attraction, recruitment and retention and greater representation/inclusion.
AI assistants in policing
Our project explored how Large Language Models (LLMs) technology represents an opportunity to introduce AI powered assistance into the prosecution process. In this project we explored the use of AI assistants, demonstrating their role in reducing labour intensive work, improving efficiency and quality while supporting training for officers.
Revolutionising digital forensics child sexual exploitation workflows through linguistics AI technology
Child sexual exploitation (CSE) investigations are increasing, with the NCA predicting a 20% growth in online CSE referrals over 23/24. We aim to speed up the identification of grooming behaviour and reduce workload in processing cases. The proposal will revolutionize digital forensics CSE workflows through the utilization of linguistics/AI technology.
When all is tread and done: is there evidential value?
This project will investigate the potential evidential value of footwear uppers caught on camera (e.g. CCTV) as a form of identification. It will investigate factors that influence the interpretation of class and individual characteristics (e.g. brand and damage, respectively) that could influence the evidential outcome of a comparison.
Assessing the organic destruction of criminally produced vegetation seized by the police.
This project is exploring the disposal of cannabis plants seized from illegal drug production in a more sustainable and cost-effective way. This is innovative and untested in the UK, however, US researchers have shown that composting cannabis can destroy Delta-9-tetrahydrocannabinol (THC) to legal levels and produce a safe usable compost.
Online lifecycle and account management
This innovative project will establish evidence-based principles for a solution that supports policing in creating, organising, managing and disposing covert personas for internet investigation and intelligence purposes. Resulting in opportunities to shape and disseminate national policy, guidance, training and create a framework to inform the future development of technical solutions.
Profile Unmasking and Linking Software Engine (PULSE)
PULSE aims to enhance digital law enforcement by identifying concealed online identities. PULSE addresses the gap in current methods of online identity verification and crime detection, which are manual, time-consuming, and limited. Leveraging advanced technologies like AI, PULSE will automate and improve the process of unmasking individuals behind pseudonymous profiles.
Transparent LLM pipeline
We aim to deliver safe, explainable, and auditable data-based decision making, including significant automation and the ability for systems to autonomously and transparently follow lines of inquiry that support investigators and analysts reach insights. This project is pushing the state of the art for the use of LLMs whilst ensuring transparency and accountability of decision making. The project builds upon significant previous work and Government-owned IP to develop and evaluate a novel AI system (called Amulet) for natural language inquiries, with transparency and auditability.
Download all 2025/2026 project summaries
When all is tread and finally done: operationalising the use of footwear uppers in casework
This project will create a machine learning tool, and develop resources for national implementation, to enable police forces to operationalise the use of footwear upper evidence. It will also investigate key areas including the influence of bias on interpretations and combining footwear mark and upper conclusions for use in court.
Pilot-testing multiple interventions to enhance mental health and wellbeing of police and law enforcement staff
This pioneering project aims to evaluate the efficacy of cost-effective evidence-based interventions in enhancing the mental wellbeing of secondary investigators and analysts (SIAs) working indirectly with traumatic material. Findings will inform and influence national policies and practices, improving support for SIAs and benefiting the broader criminal justice system.
West Yorkshire Police Trauma Tracker
The Trauma Tracker identifies attendance at the most impactful policing incidents, prompting a structured debrief to facilitate early wellbeing intervention. It tracks the impact of interventions on absence and attrition. The tracker compares against a range of datasets to establish workplace experience and the best ways to support the workforce.
STARIST: Stalking Threat AI Recognition (and) Identification Support Tool
This project will improve public safety through the development of an artificial intelligence (AI) tool, improving stalking identification during a victim’s initial police contact. This innovative, multi-agency collaboration will utilise existing stalking expertise in policing, and third sector, whilst involving proven AI technology developers. The resulting product will replicate nationally.
Generative AI for matching engine and temporal analysis in investigations
This project will create and evaluate a Gen AI multimodal matching engine enabling policing to match disparate data (sketches, text, video, and audio) and generate contextual analysis and temporal sequencing to support high workload and priority investigations.
Project DAVVIS – Digital and Visual Vehicle Identification System
Project DAVVIS aims to enhance our current national ANPR capability by augmenting basic ANPR with other associated vehicle data. The project will build on previous activity to cement ANPR as an integral tool to detect and disrupt criminal activity at a force, regional and national level.
BEAT (Baseline Ethical Assessment Tool)
BEAT (Baseline Ethical Assessment Tool) is an innovative investigative system that leverages adaptive categorisation, predictive analytics, and an ethical governance framework. Designed to address complex digital crimes, BEAT enhances law enforcement efficiency, scalability, and accountability, ensuring ethical, transparent, and effective investigative processes for modern policing challenges.
Gendering police custody: examining women’s experiences of police custody in West Yorkshire Police
The Baird Report highlighted concerns about gendered police custody. Currently little is known about the applicability of those findings beyond GMP. Through a more systematic qualitative and quantitative analysis, this research investigates women’s lived experiences of police custody in West Yorkshire, identifying challenges and best practice solutions.
Evaluating the potential use of drone-mounted multispectral imaging sensors to locate submerged human remains
The project will evaluate all available international research pertaining to the use of multispectral imaging to identify anomalies within a watercourse, specifically in the context of locating submerged human remains. The academic literary review and report produced will inform a decision to move to operational testing of the technology.
Project Nebula: harnessing image and video generative AI in law enforcement
This innovative project in partnership with academia and industry partners will enable law enforcement to harness the potential of image and video generative AI to support online intelligence and covert policing missions. It will develop skills and technical processes to apply this evolving capability within an ethical and legal framework.
Use of microbial forensics to explore the visualisation of fingerprints on human skin
Applying advanced microbial and visual enhancement technologies we will explore the feasibility of consistently visualising fingerprints on skin, assisting major crime investigations. Advancements in microbial characterisation potentially place forensic sciences at the cusp of new opportunities enabling identification and proof of physical (finger-on-skin) contact between individuals; an historically unresolved challenge.
Trustworthy skies: Co-designing a framework for ethical and transparent communication and engagement for next-generation drone use in policing
This project will work with communities across the UK to understand expectations, concerns and conditions of support for police drone use. Through independent research and citizen engagement, it will develop practical guidance, transparency tools and communication approaches to help policing use next-generation drones responsibly while maintaining trust and confidence nationally.
Considering the impact of technology and AI on police call handlers
This project examines how AI tools in police call centres affect staff wellbeing. Using the latest causal inference methods to separate cause from correlation, we will track stress, satisfaction, and absence. Results will guide workforce planning, ensuring technology strengthens resilience and service quality rather than unintentionally worsening workforce pressures.
Developing Trustworthy Bayesian Analytics for Risk Assessment in Missing Children Incidents
We will develop an evidence-based risk assessment model that assists police officers in evaluating which reports of missing children require immediate police intervention. Using a novel statistical approach, we will identify combinations of static and dynamic risk factors that indicate if a child will be exposed to harm whilst missing.
Mobile Cross-Polarization Photogrammetry Rig to Record Injuries to Aid Identification of Potential Weapons and/or Suspects
This project aims to design and build a mobile photogrammetry rig with multiple cameras and cross-polarised lights to use outside the lab. It seeks to establish a standardized method for capturing accurate 3D images of injuries, ensuring that high-fidelity records can be reliably reproduced for expert analysis and courtroom presentation.
Policing Systems to Support LGBT Victims of Hate Crime: Understanding Challenges and Developing Best Practice
LGBT victims of hate crime report negative experiences of the police. This project uses innovative social science methods to produce evidence-based understanding of why this is the case, and develop best-practice guidance and bespoke training to improve police response to LGBT hate crimes and improve relations with the LGBT community.
AI-Enhanced Behavioural Mapping for VAWG Prevention and Policing
We will use AI to analyse behavioural and contextual indicators of violence against women and girls (VAWG), integrating expert psychological insights and frontline officer experience. Using learning across multiple crime categories, we will build an evidence-based, data-led approach to improve risk assessments, interventions and outcomes, to enhance VAWG policing.
Developing a risk assessment framework for teenage relationship abuse
This project develops an evidence base on risk in Teenage Relationship Abuse (TRA) to inform early identification across services working with children and young people. It combines lived-experience and practitioner insight with advanced data analytics to identify risk and protective factors and improve understanding of how TRA manifests.
Evidence-Driven Exhibit Triage (EVIDENT)
This research investigates how a data-science driven framework to assessing digital forensic submissions could be developed to support smarter decisions on acceptance and prioritisation. Leading to the surfacing of actionable or safeguarding information sooner, early guilty pleas, and reduced CJS backlogs.
INFORM- Integrity Framework for Operational Reform and Misconduct. From Data to Decision Support
This project proposes to create an ethically governed, data-driven integrity framework that strengthens professional standards governance, supports early intervention, and underpins legitimacy, workforce confidence and public trust – all in an operationally meaningful way, developed on internal data representing ~15% of the police workforce nationally.
TACT (Talk to ACTion): Understanding and improving police communication from 999 call to CAD logs
This project investigates how police call-handlers transform spoken 999 phone calls into written Computer-Aided Dispatch (CAD) logs that guide frontline response. Analysing real emergency calls will reveal how call-handlers manage complex communication demands, generating evidence to improve training, system design, and the accuracy of operational data in Force Control Rooms.
Enhancing Stop & Search scrutiny through computer vision analysis of Body Worn Video
This project develops the technical infrastructure needed to transform Body Worn Video into structured data. Using computer vision and data science, the system will allow footage to be analysed at scale, creating new opportunities for research, oversight, and learning about how Stop and Search encounters are conducted in practice.
Forensic audio deepfake detection – developing a combined human and machine approach
Utilising a unique dataset of real and deepfake voices, this project aims to develop robust forensic audio deepfake detection methods, combining novel phonetic analysis and signal-processing research with machine learning-based deepfake detection outputs. Interpretability, explainability and sustainability will be key output goals.